• Home
  • Line#
  • Scopes#
  • Navigate#
  • Raw
  • Download
1 // Ceres Solver - A fast non-linear least squares minimizer
2 // Copyright 2012 Google Inc. All rights reserved.
3 // http://code.google.com/p/ceres-solver/
4 //
5 // Redistribution and use in source and binary forms, with or without
6 // modification, are permitted provided that the following conditions are met:
7 //
8 // * Redistributions of source code must retain the above copyright notice,
9 //   this list of conditions and the following disclaimer.
10 // * Redistributions in binary form must reproduce the above copyright notice,
11 //   this list of conditions and the following disclaimer in the documentation
12 //   and/or other materials provided with the distribution.
13 // * Neither the name of Google Inc. nor the names of its contributors may be
14 //   used to endorse or promote products derived from this software without
15 //   specific prior written permission.
16 //
17 // THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
18 // AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
19 // IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
20 // ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE
21 // LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
22 // CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
23 // SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
24 // INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
25 // CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
26 // ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
27 // POSSIBILITY OF SUCH DAMAGE.
28 //
29 // Author: sameeragarwal@google.com (Sameer Agarwal)
30 
31 #ifndef CERES_NO_LINE_SEARCH_MINIMIZER
32 
33 #include "ceres/line_search_direction.h"
34 #include "ceres/line_search_minimizer.h"
35 #include "ceres/low_rank_inverse_hessian.h"
36 #include "ceres/internal/eigen.h"
37 #include "glog/logging.h"
38 
39 namespace ceres {
40 namespace internal {
41 
42 class SteepestDescent : public LineSearchDirection {
43  public:
~SteepestDescent()44   virtual ~SteepestDescent() {}
NextDirection(const LineSearchMinimizer::State & previous,const LineSearchMinimizer::State & current,Vector * search_direction)45   bool NextDirection(const LineSearchMinimizer::State& previous,
46                      const LineSearchMinimizer::State& current,
47                      Vector* search_direction) {
48     *search_direction = -current.gradient;
49     return true;
50   }
51 };
52 
53 class NonlinearConjugateGradient : public LineSearchDirection {
54  public:
NonlinearConjugateGradient(const NonlinearConjugateGradientType type,const double function_tolerance)55   NonlinearConjugateGradient(const NonlinearConjugateGradientType type,
56                              const double function_tolerance)
57       : type_(type),
58         function_tolerance_(function_tolerance) {
59   }
60 
NextDirection(const LineSearchMinimizer::State & previous,const LineSearchMinimizer::State & current,Vector * search_direction)61   bool NextDirection(const LineSearchMinimizer::State& previous,
62                      const LineSearchMinimizer::State& current,
63                      Vector* search_direction) {
64     double beta = 0.0;
65     Vector gradient_change;
66     switch (type_) {
67       case FLETCHER_REEVES:
68         beta = current.gradient_squared_norm / previous.gradient_squared_norm;
69         break;
70       case POLAK_RIBIRERE:
71         gradient_change = current.gradient - previous.gradient;
72         beta = (current.gradient.dot(gradient_change) /
73                 previous.gradient_squared_norm);
74         break;
75       case HESTENES_STIEFEL:
76         gradient_change = current.gradient - previous.gradient;
77         beta =  (current.gradient.dot(gradient_change) /
78                  previous.search_direction.dot(gradient_change));
79         break;
80       default:
81         LOG(FATAL) << "Unknown nonlinear conjugate gradient type: " << type_;
82     }
83 
84     *search_direction =  -current.gradient + beta * previous.search_direction;
85     const double directional_derivative =
86         current.gradient.dot(*search_direction);
87     if (directional_derivative > -function_tolerance_) {
88       LOG(WARNING) << "Restarting non-linear conjugate gradients: "
89                    << directional_derivative;
90       *search_direction = -current.gradient;
91     };
92 
93     return true;
94   }
95 
96  private:
97   const NonlinearConjugateGradientType type_;
98   const double function_tolerance_;
99 };
100 
101 class LBFGS : public LineSearchDirection {
102  public:
LBFGS(const int num_parameters,const int max_lbfgs_rank,const bool use_approximate_eigenvalue_bfgs_scaling)103   LBFGS(const int num_parameters,
104         const int max_lbfgs_rank,
105         const bool use_approximate_eigenvalue_bfgs_scaling)
106       : low_rank_inverse_hessian_(num_parameters,
107                                   max_lbfgs_rank,
108                                   use_approximate_eigenvalue_bfgs_scaling),
109         is_positive_definite_(true) {}
110 
~LBFGS()111   virtual ~LBFGS() {}
112 
NextDirection(const LineSearchMinimizer::State & previous,const LineSearchMinimizer::State & current,Vector * search_direction)113   bool NextDirection(const LineSearchMinimizer::State& previous,
114                      const LineSearchMinimizer::State& current,
115                      Vector* search_direction) {
116     CHECK(is_positive_definite_)
117         << "Ceres bug: NextDirection() called on L-BFGS after inverse Hessian "
118         << "approximation has become indefinite, please contact the "
119         << "developers!";
120 
121     low_rank_inverse_hessian_.Update(
122         previous.search_direction * previous.step_size,
123         current.gradient - previous.gradient);
124     search_direction->setZero();
125     low_rank_inverse_hessian_.RightMultiply(current.gradient.data(),
126                                             search_direction->data());
127     *search_direction *= -1.0;
128 
129     if (search_direction->dot(current.gradient) >= 0.0) {
130       LOG(WARNING) << "Numerical failure in L-BFGS update: inverse Hessian "
131                    << "approximation is not positive definite, and thus "
132                    << "initial gradient for search direction is positive: "
133                    << search_direction->dot(current.gradient);
134       is_positive_definite_ = false;
135       return false;
136     }
137 
138     return true;
139   }
140 
141  private:
142   LowRankInverseHessian low_rank_inverse_hessian_;
143   bool is_positive_definite_;
144 };
145 
146 class BFGS : public LineSearchDirection {
147  public:
BFGS(const int num_parameters,const bool use_approximate_eigenvalue_scaling)148   BFGS(const int num_parameters,
149        const bool use_approximate_eigenvalue_scaling)
150       : num_parameters_(num_parameters),
151         use_approximate_eigenvalue_scaling_(use_approximate_eigenvalue_scaling),
152         initialized_(false),
153         is_positive_definite_(true) {
154     LOG_IF(WARNING, num_parameters_ >= 1e3)
155         << "BFGS line search being created with: " << num_parameters_
156         << " parameters, this will allocate a dense approximate inverse Hessian"
157         << " of size: " << num_parameters_ << " x " << num_parameters_
158         << ", consider using the L-BFGS memory-efficient line search direction "
159         << "instead.";
160     // Construct inverse_hessian_ after logging warning about size s.t. if the
161     // allocation crashes us, the log will highlight what the issue likely was.
162     inverse_hessian_ = Matrix::Identity(num_parameters, num_parameters);
163   }
164 
~BFGS()165   virtual ~BFGS() {}
166 
NextDirection(const LineSearchMinimizer::State & previous,const LineSearchMinimizer::State & current,Vector * search_direction)167   bool NextDirection(const LineSearchMinimizer::State& previous,
168                      const LineSearchMinimizer::State& current,
169                      Vector* search_direction) {
170     CHECK(is_positive_definite_)
171         << "Ceres bug: NextDirection() called on BFGS after inverse Hessian "
172         << "approximation has become indefinite, please contact the "
173         << "developers!";
174 
175     const Vector delta_x = previous.search_direction * previous.step_size;
176     const Vector delta_gradient = current.gradient - previous.gradient;
177     const double delta_x_dot_delta_gradient = delta_x.dot(delta_gradient);
178 
179     if (delta_x_dot_delta_gradient <= 1e-10) {
180       VLOG(2) << "Skipping BFGS Update, delta_x_dot_delta_gradient too "
181               << "small: " << delta_x_dot_delta_gradient;
182     } else {
183       // Update dense inverse Hessian approximation.
184 
185       if (!initialized_ && use_approximate_eigenvalue_scaling_) {
186         // Rescale the initial inverse Hessian approximation (H_0) to be
187         // iteratively updated so that it is of similar 'size' to the true
188         // inverse Hessian at the start point.  As shown in [1]:
189         //
190         //   \gamma = (delta_gradient_{0}' * delta_x_{0}) /
191         //            (delta_gradient_{0}' * delta_gradient_{0})
192         //
193         // Satisfies:
194         //
195         //   (1 / \lambda_m) <= \gamma <= (1 / \lambda_1)
196         //
197         // Where \lambda_1 & \lambda_m are the smallest and largest eigenvalues
198         // of the true initial Hessian (not the inverse) respectively. Thus,
199         // \gamma is an approximate eigenvalue of the true inverse Hessian, and
200         // choosing: H_0 = I * \gamma will yield a starting point that has a
201         // similar scale to the true inverse Hessian.  This technique is widely
202         // reported to often improve convergence, however this is not
203         // universally true, particularly if there are errors in the initial
204         // gradients, or if there are significant differences in the sensitivity
205         // of the problem to the parameters (i.e. the range of the magnitudes of
206         // the components of the gradient is large).
207         //
208         // The original origin of this rescaling trick is somewhat unclear, the
209         // earliest reference appears to be Oren [1], however it is widely
210         // discussed without specific attributation in various texts including
211         // [2] (p143).
212         //
213         // [1] Oren S.S., Self-scaling variable metric (SSVM) algorithms
214         //     Part II: Implementation and experiments, Management Science,
215         //     20(5), 863-874, 1974.
216         // [2] Nocedal J., Wright S., Numerical Optimization, Springer, 1999.
217         inverse_hessian_ *=
218             delta_x_dot_delta_gradient / delta_gradient.dot(delta_gradient);
219       }
220       initialized_ = true;
221 
222       // Efficient O(num_parameters^2) BFGS update [2].
223       //
224       // Starting from dense BFGS update detailed in Nocedal [2] p140/177 and
225       // using: y_k = delta_gradient, s_k = delta_x:
226       //
227       //   \rho_k = 1.0 / (s_k' * y_k)
228       //   V_k = I - \rho_k * y_k * s_k'
229       //   H_k = (V_k' * H_{k-1} * V_k) + (\rho_k * s_k * s_k')
230       //
231       // This update involves matrix, matrix products which naively O(N^3),
232       // however we can exploit our knowledge that H_k is positive definite
233       // and thus by defn. symmetric to reduce the cost of the update:
234       //
235       // Expanding the update above yields:
236       //
237       //   H_k = H_{k-1} +
238       //         \rho_k * ( (1.0 + \rho_k * y_k' * H_k * y_k) * s_k * s_k' -
239       //                    (s_k * y_k' * H_k + H_k * y_k * s_k') )
240       //
241       // Using: A = (s_k * y_k' * H_k), and the knowledge that H_k = H_k', the
242       // last term simplifies to (A + A'). Note that although A is not symmetric
243       // (A + A') is symmetric. For ease of construction we also define
244       // B = (1 + \rho_k * y_k' * H_k * y_k) * s_k * s_k', which is by defn
245       // symmetric due to construction from: s_k * s_k'.
246       //
247       // Now we can write the BFGS update as:
248       //
249       //   H_k = H_{k-1} + \rho_k * (B - (A + A'))
250 
251       // For efficiency, as H_k is by defn. symmetric, we will only maintain the
252       // *lower* triangle of H_k (and all intermediary terms).
253 
254       const double rho_k = 1.0 / delta_x_dot_delta_gradient;
255 
256       // Calculate: A = s_k * y_k' * H_k
257       Matrix A = delta_x * (delta_gradient.transpose() *
258                             inverse_hessian_.selfadjointView<Eigen::Lower>());
259 
260       // Calculate scalar: (1 + \rho_k * y_k' * H_k * y_k)
261       const double delta_x_times_delta_x_transpose_scale_factor =
262           (1.0 + (rho_k * delta_gradient.transpose() *
263                   inverse_hessian_.selfadjointView<Eigen::Lower>() *
264                   delta_gradient));
265       // Calculate: B = (1 + \rho_k * y_k' * H_k * y_k) * s_k * s_k'
266       Matrix B = Matrix::Zero(num_parameters_, num_parameters_);
267       B.selfadjointView<Eigen::Lower>().
268           rankUpdate(delta_x, delta_x_times_delta_x_transpose_scale_factor);
269 
270       // Finally, update inverse Hessian approximation according to:
271       // H_k = H_{k-1} + \rho_k * (B - (A + A')).  Note that (A + A') is
272       // symmetric, even though A is not.
273       inverse_hessian_.triangularView<Eigen::Lower>() +=
274           rho_k * (B - A - A.transpose());
275     }
276 
277     *search_direction =
278         inverse_hessian_.selfadjointView<Eigen::Lower>() *
279         (-1.0 * current.gradient);
280 
281     if (search_direction->dot(current.gradient) >= 0.0) {
282       LOG(WARNING) << "Numerical failure in BFGS update: inverse Hessian "
283                    << "approximation is not positive definite, and thus "
284                    << "initial gradient for search direction is positive: "
285                    << search_direction->dot(current.gradient);
286       is_positive_definite_ = false;
287       return false;
288     }
289 
290     return true;
291   }
292 
293  private:
294   const int num_parameters_;
295   const bool use_approximate_eigenvalue_scaling_;
296   Matrix inverse_hessian_;
297   bool initialized_;
298   bool is_positive_definite_;
299 };
300 
301 LineSearchDirection*
Create(const LineSearchDirection::Options & options)302 LineSearchDirection::Create(const LineSearchDirection::Options& options) {
303   if (options.type == STEEPEST_DESCENT) {
304     return new SteepestDescent;
305   }
306 
307   if (options.type == NONLINEAR_CONJUGATE_GRADIENT) {
308     return new NonlinearConjugateGradient(
309         options.nonlinear_conjugate_gradient_type,
310         options.function_tolerance);
311   }
312 
313   if (options.type == ceres::LBFGS) {
314     return new ceres::internal::LBFGS(
315         options.num_parameters,
316         options.max_lbfgs_rank,
317         options.use_approximate_eigenvalue_bfgs_scaling);
318   }
319 
320   if (options.type == ceres::BFGS) {
321     return new ceres::internal::BFGS(
322         options.num_parameters,
323         options.use_approximate_eigenvalue_bfgs_scaling);
324   }
325 
326   LOG(ERROR) << "Unknown line search direction type: " << options.type;
327   return NULL;
328 }
329 
330 }  // namespace internal
331 }  // namespace ceres
332 
333 #endif  // CERES_NO_LINE_SEARCH_MINIMIZER
334